Soccer Analytics

# import Python libraries

import pandas as pd
from pandas import read_csv
import plotly.graph_objects as go
import plotly.express as px

from IPython.display import YouTubeVideo

Professional analytics: UEFA Champions League

Event data (Atletico Madrid)
Image source: Friends of Tracking GitHub	Image source: Friends of Tracking YouTube

Example: anonymized professional soccer event data

Data source: https://github.com/metrica-sports/sample-data

df_soccer = pd.read_csv("https://raw.githubusercontent.com/metrica-sports/sample-data/master/data/Sample_Game_1/Sample_Game_1_RawEventsData.csv");
df_soccer

	Team	Type	Subtype	Period	Start Frame	Start Time [s]	End Frame	End Time [s]	From	To	Start X	Start Y	End X	End Y
0	Away	SET PIECE	KICK OFF	1	1	0.04	0	0.00	Player19	NaN	NaN	NaN	NaN	NaN
1	Away	PASS	NaN	1	1	0.04	3	0.12	Player19	Player21	0.45	0.39	0.55	0.43
2	Away	PASS	NaN	1	3	0.12	17	0.68	Player21	Player15	0.55	0.43	0.58	0.21
3	Away	PASS	NaN	1	45	1.80	61	2.44	Player15	Player19	0.55	0.19	0.45	0.31
4	Away	PASS	NaN	1	77	3.08	96	3.84	Player19	Player21	0.45	0.32	0.49	0.47
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
1740	Home	PASS	NaN	2	143361	5734.44	143483	5739.32	Player12	Player13	0.60	0.33	0.19	0.95
1741	Home	PASS	NaN	2	143578	5743.12	143593	5743.72	Player13	Player4	0.09	0.88	0.14	0.69
1742	Home	BALL LOST	INTERCEPTION	2	143598	5743.92	143618	5744.72	Player4	NaN	0.13	0.69	0.07	0.61
1743	Away	RECOVERY	BLOCKED	2	143617	5744.68	143617	5744.68	Player16	NaN	0.05	0.62	NaN	NaN
1744	Away	BALL OUT	NaN	2	143622	5744.88	143630	5745.20	Player16	NaN	0.05	0.63	0.03	1.01

1745 rows × 14 columns

px.scatter(df_soccer, x='Start X', y='Start Y', color='Type', hover_data=['Subtype'], title='All events during the game by on-field location')

Count = df_soccer.groupby("Type").size().reset_index(name="Count")
Count.sort_values('Count', ascending=False)

	Type	Count
5	PASS	799
6	RECOVERY	278
0	BALL LOST	257
3	CHALLENGE	233
7	SET PIECE	77
1	BALL OUT	51
8	SHOT	24
4	FAULT RECEIVED	22
2	CARD	4

events_df = px.bar(Count, x='Type', y='Count', color="Type", title='Total counts for types of events')
events_df.update_layout(xaxis={'categoryorder':'total descending'})

px.scatter(df_soccer, x='Start Time [s]', y='Type', hover_data=['Subtype'], color='Team', title="Timeline of events by type for Away and Home")

px.scatter(df_soccer.loc[lambda df: (df['Team'] == 'Home') & (df['From'] == 'Player10'), :], x='Start Time [s]', y='Type', color='Type', title='Timeline of events by type for player 10 on Home team')

px.scatter(df_soccer.loc[lambda df: (df['Team'] == 'Home') & (df['Period'] == 2) & (df['Type'] == 'PASS'), :], x='Start X', y='Start Y', size='Period', size_max=6, title='Home team passes, second half')

Inputting UEFA Champions League data to create graphs with Python

Displaying data using graphs from plotly

%%html
<iframe width="1024" height="600" src="https://plotly.com/python/" frameborder="0"</iframe>

labels = ['left foot', 'right foot', 'header', 'other']
values = [4, 11, 5, 0]
px.bar(x=labels, y=values, title='Bayern: Type of goal')

Inputting data from a table to create a double bar graph

%%html 
<h3>🔎 Comparing teams: Do similar techniques lead to goal scoring?</h3><br><iframe width="1024" height="400" src="https://www.uefa.com/uefachampionsleague/season=2021/statistics/round=2001252/clubs/kind=goaltypes/index.html" frameborder="0" ></iframe>

🔎 Comparing teams: Do similar techniques lead to goal scoring?

goal_type=['left foot', 'right foot', 'header', 'other']

bar_goal_type = go.Figure(data=[
    go.Bar(name='Bayern', x=goal_type, y=[4, 11, 5, 0]),
    go.Bar(name='Barcelona', x=goal_type, y=[6, 10, 1, 1])])

bar_goal_type.update_layout(title_text='Bayern versus Barcelona: Type of goal')

Inputting data to create a circle graph

%%html
<h3>🔎 How do on target goals and unsuccessful goal attempts compare?</h3><br><iframe width="1024" height="400" src="https://www.uefa.com/uefachampionsleague/season=2021/statistics/round=2001252/clubs/kind=attempts/index.html" frameborder="0" ></iframe>

🔎 How do on target goals and unsuccessful goal attempts compare?

px.pie(names=['On target', 'Off target', 'Blocked', 'Against woodwork'], 
       values=[48, 38, 22, 4], 
       title='Barcelona: Goal attempts')

🔎 How do ball possession and scoring relate?

Data inputted in csv file on March 10, 2021, prior to knock-outs.

%%writefile possession.csv
Passes completed (%),Average ball possession (%),Goal difference,Average scored per game,Team
83,50,-8,0.5,Rennes
82,50,-8,0.33,Olympiacos
82,47,-9,0.67,Zenit
86,45,-9,0.67,Dynamo_Kyiv
77,48,-7,1.67,Salzburg
80,45,-11,1.17,Istanbul_Basaksehir
76,40,7,1.75,Porto
73,46,-9,0.67,Midtjylland
75,40,-5,1,Krasnodar
77,42,-5,0.83,Lokomotiv_Moskva
91,56,8,2.25,Barcelona
89,60,3,1.71,Real_Madrid
88,59,14,2.14,Man_City
86,51,8,2.13,Dortmund
86,61,16,3.14,Bayern
88,54,10,2.25,Juventus
85,54,13,2.14,Chelsea
86,61,0,1.63,Sevilla
82,53,9,1.75,Liverpool
87,53,10,2.25,Paris
80,51,-3,1.38,Leipzig
81,48,-2,1,Atletico
84,43,5,2.29,Monchengladbach
84,46,1,1.71,Lazio
80,45,1,1.43,Atalanta
81,55,0,1.17,Ajax
85,51,-2,1.17,Internazionale
82,51,-2,1.33,Club_Brugge
84,50,5,2.5,Man_United
86,47,-7,0.83,Shakhtar_Donetsk
85,47,-12,0.83,Ferencvaros
84,50,-11,0.33,Marseille

Overwriting possession.csv

possession_df = pd.read_csv('possession.csv')
possession_df.sort_values('Goal difference', ascending=False)

	Passes completed (%)	Average ball possession (%)	Goal difference	Average scored per game	Team
14	86	61	16	3.14	Bayern
12	88	59	14	2.14	Man_City
16	85	54	13	2.14	Chelsea
19	87	53	10	2.25	Paris
15	88	54	10	2.25	Juventus
18	82	53	9	1.75	Liverpool
13	86	51	8	2.13	Dortmund
10	91	56	8	2.25	Barcelona
6	76	40	7	1.75	Porto
22	84	43	5	2.29	Monchengladbach
28	84	50	5	2.50	Man_United
11	89	60	3	1.71	Real_Madrid
24	80	45	1	1.43	Atalanta
23	84	46	1	1.71	Lazio
25	81	55	0	1.17	Ajax
17	86	61	0	1.63	Sevilla
26	85	51	-2	1.17	Internazionale
21	81	48	-2	1.00	Atletico
27	82	51	-2	1.33	Club_Brugge
20	80	51	-3	1.38	Leipzig
9	77	42	-5	0.83	Lokomotiv_Moskva
8	75	40	-5	1.00	Krasnodar
4	77	48	-7	1.67	Salzburg
29	86	47	-7	0.83	Shakhtar_Donetsk
0	83	50	-8	0.50	Rennes
1	82	50	-8	0.33	Olympiacos
7	73	46	-9	0.67	Midtjylland
3	86	45	-9	0.67	Dynamo_Kyiv
2	82	47	-9	0.67	Zenit
5	80	45	-11	1.17	Istanbul_Basaksehir
31	84	50	-11	0.33	Marseille
30	85	47	-12	0.83	Ferencvaros

scatter_df = px.scatter(possession_df,
                    x="Average ball possession (%)", y="Goal difference", trendline="ols",
                    title="Relationship between average ball possesion (%) and goal difference")
scatter_df.show()

scatter_df = px.scatter(possession_df,
                    x="Average ball possession (%)", y="Goal difference",
                    size="Average scored per game", color="Team",
                    title="Relationship between average ball possesion (%) and goal difference by team")
scatter_df.show()